Usage and billing

Senaiy Lab has one billing model: a prepaid wallet, spent per request. No seats, no subscriptions, no expiry on credit.

The usage API

curl "https://platform.senaiy.ai/v1/usage?from_date=2026-08-01&to_date=2026-08-31" \
  -H "Authorization: Bearer sk-sf-..."

Every metered request appears with its model, endpoint, token counts, outcome and rated cost in USD — the same numbers the wallet is debited with. Agent and workflow runs attribute their inner model calls here too, so the ledger always adds up.

How money flows

1. Top up the wallet in the portal (card via Stripe, or a manual finance-approved top-up). Credit is applied the moment payment completes.
2. Usage is deducted per request; the portal's Usage & costs page shows live analytics and a per-request record with CSV export.
3. Auto-reload (optional): when the balance falls to your threshold, the saved card tops it back up to your target — with a monthly cap you control. Work never stops mid-task.
4. When the balance is exhausted (past a small negative tolerance), the gateway refuses further calls until the next top-up — you can never silently overspend.

Rate limits

Requests beyond your plan's rate limits return 429 — back off and retry. Concurrency caps for agents queue rather than fail where possible.